Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS7023371 B2
Publication typeGrant
Application numberUS 10/920,500
Publication dateApr 4, 2006
Filing dateAug 18, 2004
Priority dateApr 12, 2004
Fee statusPaid
Also published asUS7224304, US7501969, US20050225472, US20060139197, US20070222655
Publication number10920500, 920500, US 7023371 B2, US 7023371B2, US-B2-7023371, US7023371 B2, US7023371B2
InventorsBrian F. Schoner
Original AssigneeBroadcom Corporation
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Method and apparatus for an image canceling digital-to-analog converter (DAC)
US 7023371 B2
Abstract
A method and apparatus for an image canceling digital-to-analog converter is disclosed. Up-sampling and noise shaping is used to produce a stream of digital sample values at a relatively higher sampling rate than the sampling rate of the digitized input samples, each higher sampling rate sample having fewer bits than the original samples. The higher sampling rate stream is then distributed for sequential conversion by multiple digital-to-analog converters each operating at a lower sampling rate. The outputs of the converters are then combined to form a combined output signal. Most spectral images normally observed in a standard or conventional DAC are attenuated in the combined output signal of an embodiment in accordance with the present invention. Any spectral images that remain are further from the signal of interest, permitting the use of lower cost filtering.
Images(11)
Previous page
Next page
Claims(27)
1. An image canceling digital-to-analog converter (DAC) comprising:
an up-sampler for receiving a first stream of digital values at a first sample rate, the first stream of digital values representative of a first signal, each of the digital values comprising a first number of bits, the up-sampler producing, at a second sample rate, a second stream of digital values based upon the first stream of digital values;
a clock source producing a plurality of clock signals substantially equally spaced in time;
a plurality of digital-to-analog converters, each of the plurality of digital-to-analog converters periodically accepting, under control of one of the plurality of clock signals, a digital value from the second stream of digital values, the outputs from the plurality of digital-to-analog converters forming a plurality of analog output signals;
a combiner adapted to combine the plurality of analog output signals to form a combined output signal; and
a noise shaper for adjusting selected digital values of the second stream of digital values in order to minimize spectral images in the combined output signal.
2. The converter of claim 1 wherein the first sample rate is less than the second sample rate.
3. The converter of claim 1 wherein the first number of bits is greater than the second number of bits.
4. The converter of claim 1 wherein the first stream comprises 10-bit digital values.
5. The converter of claim 1 wherein the second stream comprises 8 bit digital values.
6. The converter of claim 1 wherein the plurality of digital-to-analog converters comprises four digital-to-analog converters.
7. The converter of claim 1 wherein the combiner adds the plurality of analog output signals to form the combined output signal.
8. The converter of claim 1 wherein the noise shaper implements a four-tap moving average filter.
9. The converter of claim 1 wherein the noise shaper is adapted to minimize total error in the combined output signal.
10. A method of operating an image canceling digital-to-analog converter (DAC), the method comprising:
receiving a first stream of digitized signal samples;
up-sampling the first stream of digitized signal samples to produce a second stream of digitized signal samples;
performing noise shaping upon the second stream of digitized signal samples;
sequentially converting to analog output signals, in a round-robin fashion, each of a predetermined number of samples of the second stream of digitized signal samples;
combining the analog output signals to produce a combined output signal; and
the noise shaping adapted so as to minimize spectral images in the combined output signal.
11. The method of claim 10 wherein samples of the first stream comprise a relatively greater number of bits, and samples of the second stream comprise a relatively lesser number of bits.
12. The method of claim 11 wherein samples of the first stream comprise 10 bits.
13. The method of claim 11 wherein samples of the second stream comprise 8 bits.
14. The method of claim 10 wherein the sample rate of the first stream is a relatively lower sample rate, and the sample rate of the second stream is a relatively higher sample rate.
15. The method of claim 10 wherein the sample rate of the second stream is an integer multiple of the sample rate of the first stream, for integer values greater than or equal to 2.
16. The method of claim 10 wherein the noise shaping is adapted to minimize total error in the combined output signal.
17. The method of claim 10 wherein the noise shaping applies a 4-tap moving average filter to the samples of the second stream.
18. The method of claim 10 wherein the combining adds the analog output signals to form the combined output signal.
19. A machine-readable storage, having stored thereon a computer program having a plurality of code sections for operating an image canceling digital-to-analog converter (DAC), the code sections executable by a machine for causing the machine to perform the operations comprising:
receiving a first stream of digitized signal samples;
up-sampling the first stream of digitized signal samples to produce a second stream of digitized signal samples;
performing noise shaping upon the second stream of digitized signal samples;
sequentially converting to analog output signals, in a round-robin fashion, each of a predetermined number of samples of the second stream of digitized signal samples;
combining the analog output signals to produce a combined output signal; and
the noise shaping adapted so as to minimize spectral images in the combined output signal.
20. The machine-readable storage of claim 19 wherein samples of the first stream comprise a relatively greater number of bits, and samples of the second stream comprise a relatively lesser number of bits.
21. The machine-readable storage of claim 19 wherein samples of the first stream comprise 10 bits.
22. The machine-readable storage of claim 19 wherein samples of the second stream comprise 8 bits.
23. The machine-readable storage of claim 19 wherein the sample rate of the first stream is a relatively lower sample rate, and the sample rate of the second stream is a relatively higher sample rate.
24. The machine-readable storage of claim 19 wherein the sample rate of the second stream is an integer multiple of the sample rate of the first stream, for integer values greater than or equal to 2.
25. The machine-readable storage of claim 19 wherein the noise shaping is adapted to minimize total error in the combined output signal.
26. The method of claim 19 wherein the noise shaping applies a 4-tap moving average filter to the samples of the second stream.
27. The method of claim 19 wherein the combining adds the analog output signals to form the combined output signal.
Description
RELATED APPLICATIONS

This application makes reference to, claims priority to, and claims the benefit of U.S. Provisional Patent Application Ser. No. 60/561,332, entitled “METHOD AND APPARATUS FOR AN IMAGE CANCELING DIGITAL-TO-ANALOG (DAC)”, filed Apr. 12, 2004, the complete subject matter of which is hereby incorporated herein by reference, in its entirety.

FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[Not Applicable]

MICROFICHE/COPYRIGHT REFERENCE

[Not Applicable]

BACKGROUND OF THE INVENTION

Images are a common problem when digital-to-analog converters (DACs) are employed to reproduce analog waveforms. In general, DACs have a zero-order hold output. If samples representing a signal having a frequency, F, are applied to a zero-order hold DAC clocked at a sample frequency, Fs, the output will have spectral images at (F +/−kFs), for every integer k. FIG. 1A illustrates some of the problems associated with spectral images for a DAC of conventional design. Referring to FIG. 1A, digitized samples of a signal having a frequency F=1 MHz are applied to a zero-order hold DAC clocked at a sampling frequency (150) of Fs=8 MHz. The resulting spectral images are illustrated at (8−1) MHz (141) and (8+1) MHz (142), as well as (16−1) MHz (143) and (16+1) MHz (144). The amplitudes of the spectral images have a slow sinc roll-off (not shown). For the example of FIG. 1A, the 1 MHz sinusoidal signal sampled at an 8 MHz sampling frequency (150) has spectral images at 7 MHz (141), 9 MHz (142), 15 MHz (143), 17 MHz (144), and so on. One conventional method of removing the spectral images involves filtering the output of the DAC using a low-pass filter, as shown in FIG. 1B.

FIG. 1B illustrates suppression of the spectral images shown in FIG. 1A, by employing passband filtering of the output of a DAC of conventional design. Referring to FIG. 1B, during passband filtering, the desired 1 MHz signal (110) falls within the passband of a filter having the illustrated response curve (120), while the spectral images (141)–(148) fall within the stopband of the filter. High-quality filters with sharp cutoff frequencies are difficult to build, and are often expensive. A common way to reduce the cost of the filter is to use a higher sample clock frequency, Fs (150), in order to move the spectral images further from the signal of interest.

FIG. 1C is a diagram that illustrates filtering in a DAC of conventional design utilizing a higher clock or sampling frequency, Fs (150). Referring to FIG. 1C, filtering using the higher sample clock frequency (150) permits the use of a filter having a slower cutoff as shown by the response curve (120) illustrated in FIG. 1C. Such a response curve reduces the cost of the filter, when compared to filters having a sharper response. However, DACs that operate at the higher sampling frequency (150) are difficult to design, and are typically more expensive than those operating at lower frequencies. First-order as well as higher-order holds are possible, and are sometimes mentioned in research, but implementation is usually not practical.

Many conventional DACs are built from a number of switched current sources. There are several different arrangements for such switched current sources. A simple 10-bit DAC, for example, may have 10 binary-weighted current sources, or 1024 uniformly-weighted current sources. Switching the current sources on or off changes the output current proportional to the 10-bit digital input. Each digital value at the input is thereby converted into an analog current strength at the output. FIG. 1D is a block diagram of a conventional DAC.

Further limitations and disadvantages of conventional and traditional approaches will become apparent to one of ordinary skill in the art through comparison of such systems with the present invention as set forth in the remainder of the present application with reference to the drawings.

BRIEF SUMMARY OF THE INVENTION

Aspects of the present invention may be found in an image canceling digital-to-analog converter (DAC) comprising an up-sampler, a clock source, a plurality of digital-to-analog converters, a combiner, and a noise shaper. In an embodiment of the present invention, the up-sampler may receive a first stream of digital values at a first sample rate. The first stream of digital values may be representative of a first signal, and each of the digital values may comprise a first number of bits. The up-sampler may produce, at a second sample rate, a second stream of digital values based upon the first stream of digital values. The clock source may produce a plurality of clock signals substantially equally spaced in time. Each of the plurality of digital-to-analog converters may periodically accept, under control of one of the plurality of clock signals, a digital value from the second stream of digital values, and the outputs from the plurality of digital-to-analog converters may form a plurality of analog output signals. The combiner may be adapted to combine the plurality of analog output signals to form a combined output signal, and the noise shaper may adjust selected digital values of the second stream of digital values in order to minimize spectral images in the combined output signal.

In an embodiment in accordance with the present invention, the first sample rate may be less than the second sample rate, and the first number of bits may be greater than the second number of bits. The first stream may comprise 10-bit digital values, the second stream may comprise 8 bit digital values, and the plurality of digital-to-analog converters may comprise four digital-to-analog converters. The combiner in an embodiment of the present invention may add the plurality of analog output signals to form the combined output signal, and the noise shaper may implement a four-tap moving average filter. The noise shaper may be adapted to minimize total error in the combined output signal.

Additional aspects of the present invention may be seen in a method of operating an image canceling digital-to-analog converter (DAC). Such a method may comprise receiving a first stream of digitized signal samples, up-sampling the first stream of digitized signal samples to produce a second stream of digitized signal samples, and performing noise shaping upon the second stream of digitized signal samples. An embodiment of the present invention may also comprise sequentially converting to analog output signals, in a round-robin fashion, each of a predetermined number of samples of the second stream of digitized signal samples, and combining the analog output signals to produce a combined output signal. The noise shaping may be adapted so as to minimize spectral images in the combined output signal.

In an embodiment of the present invention, samples of the first stream may comprise a relatively greater number of bits, and samples of the second stream may comprise a relatively lesser number of bits. Samples of the first stream may comprise 10 bits, and samples of the second stream may comprise 8 bits. The sampling rate of the first stream may be a relatively lower sampling rate, and the sampling rate of the second stream may be a relatively higher sampling rate. The sampling rate of the second stream may be an integer multiple of the sampling rate of the first stream, for integer values greater than or equal to 2. The noise shaping may be adapted to minimize total error in the combined output signal, and the noise shaping may apply a 4-tap moving average filter to the samples of the second stream. The combining may add the analog output signals to form the combined output signal.

Further aspects of the present invention may be observed in a machine-readable storage, having stored thereon a computer program having a plurality of code sections executable by a machine for causing the machine to perform the operations of the method described above.

These and other features and advantages of the present invention may be appreciated from a review of the following detailed description of the present invention, along with the accompanying figures in which like reference numerals refer to like parts throughout.

BRIEF DESCRIPTION OF SEVERAL VIEWS OF THE DRAWINGS

FIG. 1A illustrates some of the problems associated with spectral images for a DAC of conventional design.

FIG. 1B illustrates suppression of the spectral images shown in FIG. 1A, by employing passband filtering of the output of a DAC of conventional design.

FIG. 1C is a diagram that illustrates filtering in a DAC of conventional design utilizing a higher clock or sampling frequency, Fs.

FIG. 1D is a block diagram of a conventional DAC.

FIG. 2 shows a graph illustrating the reproduced 1 MHz signal, the spectral images spaced symmetrically about the 16 MHz effective DAC sampling frequency, and the cancelled spectral images that are avoided using an image canceling digital-to-analog converter (DAC) in accordance with an embodiment of the present invention.

FIG. 3 is a block diagram of an exemplary image canceling DAC capable of converting 10-bit digitized samples into an analog output signal, in accordance with an embodiment of the present invention.

FIG. 4 shows a flowchart illustrating an exemplary method of operating an image canceling DAC such as, for example, the image canceling DAC of FIG. 3, in accordance with an embodiment of the present invention.

FIG. 5 shows a block diagram illustrating a mathematical model that may be used to analyze the expected behavior of an image canceling DAC, in accordance with an embodiment of the present invention.

FIG. 6 shows a graph illustrating the attenuation of the 4-tap comb filter of FIG. 5, in accordance with an embodiment of the present invention.

FIG. 7 illustrates the simulated output frequency spectrum for a standard or conventional 10-bit DAC reproducing a 60.123 MHz input signal at a sample rate of 108 MHz.

FIG. 8 illustrates the simulated output frequency spectrum for an exemplary 108 MHz 4× image canceling DAC reproducing a 60.123 MHz input signal, in accordance with an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Aspects of the present invention relate to the conversion of digital data to analog values using digital-to-analog converters. More specifically, aspects of the present invention relate to a method and apparatus for a digital-to-analog converter having image canceling properties wherein spectral images in the output signal that are related to the input waveform and the sampling rate are reduced or eliminated.

In an embodiment of the invention, the invention clocks different parts of the DAC from different phases of the clock. Noise shaping may be utilized to shape the noise between phases of the clock. This simple, low-cost technique cancels the DAC spectral images, and gives the DAC the effective performance of a DAC with a higher clock frequency as illustrated in FIG. 2. In many cases, application of an embodiment of the present invention the invention is less expensive than traditional alternatives of using a higher sample clock rate, or a sharper low-pass filter at the DAC output.

FIG. 2 shows a graph illustrating the reproduced 1 MHz signal 210, the spectral images 230 spaced symmetrically about the 16 MHz effective DAC sampling frequency 240, and the cancelled spectral images 211216 that are avoided using an image canceling digital-to-analog converter (DAC) in accordance with an embodiment of the present invention. The illustration of FIG. 2 also shows the response 220 of a filter that may be used to attenuate the remaining spectral images 230 that may be produced by an image canceling DAC in accordance with the present invention. The spectral images 230 produced by an embodiment of the present invention appear at a higher relative frequency within the spectrum shown in FIG. 2 than those of conventional DACs, as an image canceling DAC in accordance with the present invention is designed to eliminate the generation of lower-frequency spectral images. In this regard, a 4 MHz DAC with 4× image cancellation having the performance shown in the illustration of FIG. 2 has the effective performance of a 16 MHz DAC.

FIG. 3 is a block diagram of an exemplary image canceling DAC 300 capable of converting 10-bit digitized samples 305 into an analog output signal 380, in accordance with an embodiment of the present invention. In an embodiment of the present invention, a DAC such as, for example, the 10-bit DAC of FIG. 1D may be modified by segmenting the 10-bit DAC into a plurality of smaller DACs, each DAC processing less than the full number of bits of the digitized sample stream. For example, the 10-bit DAC of FIG. 1D may be segmented into four (4) 8-bit DACs. In the illustration of FIG. 3, the image canceling DAC 300 comprises a 4× up-sampler 310, a noise shaper 311, a four-phase clock source 320, four 8-bit DACs 330, 340, 350, 360, and a combiner 370. The 4× up-sampler 310 accepts 10-bit digitized input samples 305 at a sampling frequency of Fs, and generates 8-bit samples at output 312 that are passed to noise shaper 311 at a sampling frequency of 4*Fs. Although the example of FIG. 3 uses an up-sampling multiplier of four (4), the present invention is not limited in this regard and may be applied for any integer up-sampling rate greater than one (1). The 4× up-sampler 310 and noise shaper 311 process the 10-bit input samples 305 as described in greater detail, below. The output of the noise shaper 311 is then passed to the inputs of the four 8-bit DACs 330, 340, 350, 360. Operating in synchronism with the 4× up-sampler 310 and noise shaper 311, the four-phase clock source 320 produces four clock phases #0 332, #1 342, #2 352, #3 362 having the quadrature clock phase relationship shown in FIG. 3A. Each of the four clock phases #0 332, #1 342, #2 352, #3 362 is used for clocking, in a round-robin fashion, an 8-bit sample from the output 313 of the noise shaper 311 into one of the four 8-bits DACs 330, 340, 350, 360. In an embodiment of the present invention, the four 8-bit DACs 330, 340, 350, 360 may be either current or voltage output DACs. The four digital up-sampled and noise shaped values are sequentially converted by the four 8-bit DACs 330, 340, 350, 360. The outputs of the four 8-bit DACs 330, 340, 350, 360 are then passed to the combiner 370, that combines the four output signals into the combined output 380 of the image canceling DAC 300.

Each of the four 8-bit DACs 330, 340, 350, 360 possesses ¼ the range of a DAC such as, for example, the 10-bit DAC shown in FIG. 1D. In an embodiment of the present invention, the least significant bit (LSB) of each of the four 8-bit DACs 330, 340, 350, 360 may be weighted such that a 1 LSB input change results in a DAC output change that is the same as the change in the output of the conceptually equivalent 10-bit DAC due to a 1 LSB input change. Because the four 8-bit DACs 330, 340, 350, 360 are clocked from four clock phases #0 332, #1 342, #2 352, #3 362, respectively, the four DACs 330, 340, 350, 360 of the image canceling DAC of FIG. 3 have the ability to accept samples at four times the original sample rate, Fs, as produced at the output 313 of the noise shaper 311. Each of the four 8-bit DACs 330, 340, 350, 360, however, is still clocked at the original sample rate, Fs, eliminating the need to use the high speed signal conversion circuitry that might otherwise be employed in a single conventional 10-bit DAC operating at the higher 4*Fs sampling rate.

FIG. 4 shows a flowchart illustrating an exemplary method of operating an image canceling DAC such as, for example, the image canceling DAC 300 of FIG. 3, in accordance with an embodiment of the present invention. In an embodiment of the present invention, the four-times up-sampler 310 and noise shaper 311 may produce 8-bit samples for the four 8-bit DACs to be sequentially converted to an analog output signal in a round-robin fashion, according to the following method. The method of FIG. 4 begins when the number of DACs to be used in the production of the analog output signal is set to a value, K (block 410). For example, for the 10-bit image canceling DAC 300 of FIG. 3, the value of variable K may be set to 4, indicating the use of four DACs. The method then sets each of K values in an array variable, remainder, to zero (0) (block 412). The value of a variable, prevsample, is then set to zero (0) (block 414), because no previous sample has been processed.

Next, the method begins processing digitized input samples by receiving the first of a sequence of new M-bit digitized samples, at a sample rate, Fs, samples per second (block 416). To illustrate, for the 10-bit image canceling DAC 300 of FIG. 3, the value of M may be set to ten (10). The method of FIG. 4 then sets the value of a loop control variable, I, to zero (0) (block 418), in preparation for the generation and processing of K up-sampled values. A check is then made whether K up-sampled values have already been processed (block 420). If K up-sampled values have already been processed from the most recent digitized input sample, newsample, the most recent digitized input sample, newsample, is saved in a variable, prevsample, (block 422) and a new digitized input sample is retrieved from the sequence of M-bit digitized input samples (block 416). If K up-sampled values have not already been processed from the most recent M-bit digitized input sample, newsample, the next up-sample value, up-sample, is computed (block 424) using interpolation of K points between the previous input sample value, prevsample, and the new input sample value, newsample.

An initial value of the sample for the next DAC in sequence, phase[I], is then computed (block 426) by taking the integer portion of the value of the interpolated sample point divided by K. The remainder (or error) resulting from the calculation of an integer approximation of the interpolated value is then calculated and saved in remainder[I] (block 428). The method then calculates the total error, err, of the remainders for each of the previously generated K−1 interpolated points (blocks 432438). The value of variable, err, is then tested to determined whether the total error is greater than zero (0) (block 440). If the value of the total error variable, err, is greater than zero (0), the value of the sample for the next DAC in sequence, phase[I], is adjusted by increasing it by one (1) (block 442), and the value of the variable, remainder, for the interpolated sample point is decreased by one (1) (block 444). The value of the sample for the next DAC in sequence, phase[I], is then output to the respective DAC at the next 1/(4*Fs) time boundary (block 446). The value of the loop variable, I, is then incremented (block 448), and the method returns to check whether further interpolated values of the variable, up-sample, are to be processed for the most recent digitized input sample (block 420).

If, however, the value of the total error variable, err, is less than or equal to zero (0) (block 440), the value of the sample for the next DAC in sequence, phase[I], is output, without adjustment, to the respective DAC at the next 1/(4*Fs) time boundary (block 446). The value of the loop variable, I, is then incremented (block 448), and the method returns to check whether further values of the variable, up-sample, are to be processed for the most recent digitized input sample (block 420).

FIG. 5 shows a block diagram illustrating a mathematical model that may be used to analyze the expected behavior of an image canceling DAC, in accordance with an embodiment of the present invention. The arrangement of FIG. 5 is mathematically equivalent to the exemplary image canceling DAC 300 of FIG. 3, but the digital-to-analog conversion has been moved after the summation. In the arrangement of FIG. 3, the four, 8-bit DACs effectively function as a 4-tap comb filter. The illustration of FIG. 5 also allows an understanding of the image cancellation property of an embodiment of a digital-to-analog converter in accordance with the present invention. The DAC illustrated in FIG. 3 is effectively operating at a sampling frequency of 4*Fs, and will gain all of the benefits of a DAC having a higher sampling rate. That is, in an embodiment of the present invention, the spectral images will be positioned further from the original signal, and from each other, and the DAC may benefit from noise shaping techniques, described below.

FIG. 6 shows a graph illustrating the attenuation of the 4-tap comb filter 520 of FIG. 5, in accordance with an embodiment of the present invention. The 4-tap comb filter 520 functions as a low-pass filter, and attenuates the output of the DAC 530 of FIG. 5 at high frequencies. Because the original signal input 505 is sampled at Fs, it is a low frequency signal when compared to the sample clock of the DAC 530, that operates at 4*Fs. The curve 610 shows the additional roll-off from the comb filter 520 of FIG. 5, that imparts significant attenuation at high frequencies. The curve 620 shows the attenuation that is characteristic of all sample-and-hold DACs, and illustrates the 6 dB attenuation at f=0.6 Fs. The roll-off of curve 620 may be represented as:
sinc (π*f/Fs)

In an embodiment of the present invention, the digital values distributed to the K digital-to-analog converters may first be adjusted using noise shaping provided by a noise shaping function such as the noise shaper 311 shown in FIG. 3. If each of the original 10-bit digitized samples 305 is first up-sampled at a sampling rate of 4*Fs, truncated to an 8-bit value, and converted to an analog value, the resulting analog signal exhibits the performance of only an 8-bit DAC. An embodiment of the present invention, however, uses a novel application of noise shaping to recover a significant amount of the performance of a 10-bit DAC, by shaping the noise across all phases of the sampling clock. This results in an averaging of the quantization noise from, for example, each individual 8-bit DACs 330, 340, 350, 360 of FIG. 3. The noise shaping function in an embodiment of the present invention may be performed using any of a number of noise shaping arrangements such as, for example, a sigma-delta approach. The applicant has observed good performance in an N× image canceling DAC using an N-tap moving average, that will average the quantization error across each of the N smaller DACs.

Table 1, below, illustrates the operation of a 4-tap moving average filter for noise shaping in an image canceling DAC in accordance with an embodiment of the present invention. The values shown in Table 1 illustrate the values of variables used in the method diagrammed in FIG. 4, for a 10-bit (M=10) image canceling DAC employing fours (K=4) 8-bit DACs. The 10-bit input sample values in the example of Table 1 may be assumed to be a linear ramp.

TABLE 1
10-bit
input
sample at up-
Fs sample phase[I] remainder output = sum(phase[0–3]) ideal output
800 800 200 0/4
825 206 1/4
850 213 −2/4  
(212 + 1) (2/4 − 1)
875 218 3/4 200 + 206 + 213 + 218 = 837 837.5 (+)
900 900 226 −4/4   206 + 213 + 218 + 226 = 863 862.5 (−)
(225 + 1) (0/4 − 1)
925 231 1/4 213 + 218 + 226 + 231 = 888 887.5 (−)
950 237 2/4 218 + 226 + 231 + 237 = 912 912.5 (+)
975 243 3/4 226 + 231 + 237 + 243 = 937 937.5 (+)

As illustrated in the method of FIG. 4 and the values in the columns of Table 1, in an embodiment of the present invention, the original 10-bit digitized samples 305 may be interpolated (e.g., in FIG. 3 and Table 1, by a factor of 4×) to create an up-sampled linear ramp [e.g., 800, 825, 875, etc.]. The up-sampled values then undergo noise shaping, and are distributed to a number of smaller DACs. In the examples of FIG. 3 and Table 4, the number of DAC used is four (i.e., 8-bit DACs 330, 340, 350, 360). For example, the first 10-bit input sample value [800] is sent to the first DAC as [800/4=200+0/4.] Each time a value is sent to a DAC, the remainder of the division is fed back to the next three calculations. For example, in the third row of Table 1, the value [850/4=212+2/4] “sees” the remainders [0/4, 1/4] from previous two calculations. The sum of these earlier greater than zero, and results in the value, phase[I], being rounded up to 213 [850/4=212+1], and the original remainder value being adjusted by decrementing it to −2/4 [−2/4=2/4−1].

To illustrate another example, in row five of Table 1, the value 225 [225=900/4] with a remainder of +0/4, “sees” the remainders [1/4, −2/4, 3/4] from the prior three calculations. Because the sum of the remainders is less than or equal to zero, the value, phase[I], [225] is not rounded up, and the remainder is not decremented.

In an embodiment in accordance with the present invention, such as that shown in FIG. 3 and Table 1, the phases of an image canceling DAC cooperate in this manner to minimize the total error at the output. To illustrate, the two rightmost columns of Table 1 show the sum of the outputs of the four 8-bit DACs, and the “ideal” output. Note that the output of the four DACs in the example is always within ½ LSB of the ideal output, and tends to oscillate +1/2, −1/2 in an attempt to have zero average error.

The applicant has performed simulations showing that an image canceling DAC according to an embodiment of the present invention exhibits many of the properties of a DAC clocked at N times that sample rate of the original digitized input values. For example, low-frequency spectral images are cancelled, and the signal-to-noise (SNR) is as good as a traditional DAC. An image canceling DAC in accordance with an embodiment of the present invention has the same sinc roll-off as a DAC clocked at N times the original, lower sample rate. Such a DAC also has a second roll-off due to the summation of the different phases from the individual smaller DACs. This second roll-off is the same as the roll-off from an N-tap moving average (or comb) filter. An image canceling DAC in accordance with an embodiment of the present invention may permit the reproduction of signals at frequencies beyond the Nyquist limit of Fs/2. The Nyquist limit for an N× image cancellation DAC is N*Fs/2.

The image cancellation technique of an embodiment of the present invention may be applied to almost any digital-to-analog converter application. Although the above description provided examples that utilized 4× image cancellation on a 10-bit current DAC, the present invention is not limited in this regard. An embodiment in accordance with the present invention may have application in almost any system in which digital-to-analog conversion is used. Accordingly, the present invention is not limited, for example, to 10-bit DACs, or current DACs, or DACs built from uniformly weighted current sources.

FIG. 7 illustrates the simulated output frequency spectrum for a standard or conventional 10-bit DAC reproducing a 60.123 MHz input signal 710 at a sample rate of 108 MHz. Each output is repeated for 4 clock cycles to show the images. Because the 60.123 MHz input signal is beyond the Nyquist limit, there is a spectral image 720 at approximately 108−60=48 MHz. Additional spectral images 730, 740 also exist at roughly 108+60=168 MHz, and 216−60=156 MHz, respectively. The illustration of FIG. 7 shows a frequency range of approximately zero (0) MHz to 216 MHz.

FIG. 8 illustrates the simulated output frequency spectrum for an exemplary 108 MHz 4× image canceling DAC reproducing a 60.123 MHz input signal 810, in accordance with an embodiment of the present invention. Each of the four DAC sections of the simulated image canceling DAC of FIG. 8 is limited to 256 levels (8 bits). Note that no spectral images are present—even the spectral image at 108−60=48 MHz is suppressed. The noise floor and signal amplitude of the image canceling DAC of FIG. 8 are both comparable to a standard 10-bit DAC, such as for the standard or conventional DAC shown in FIG. 1D.

The following C-language code example may be utilized to illustrate the operation of a 4× image canceling DAC such as, for example, the image canceling DAC of FIG. 3, in accordance with an embodiment of the present invention.

// C model of 4x anti-alias DAC
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
// set SCALE to 1 for a normal DAC, 4 for anti-alias DAC
#define SCALE 4
int main(int argc, char *argv[ ]) {
double freq, dither, in, dsum;
int out, x, y, phase, sum;
int dacs[4] = {0,0,0,0};
double err[4] = {0,0,0,0};
// 60.123 MHz input signal, sampled at 108MHz
freq = 3.14159 * 2.0 * 60123.0 / 108000.0;
out = 0;
phase = 0;
dither = 0.0;
// Effectively oversample by dividing frequency by SCALE
freq /= SCALE;
// Run through 25000 samples
for (x = 0; x < 25000*SCALE; x++) {
in = sin (x * freq) * 480 + 512; // 10-bit DAC range [0,1023]
dither = (double) (rand( ));
dither = dither / 32768.0 − 0.5;
// sum errors from all other phases
err[phase] = 0.0;
dsum = 0.0;
for (y = 0; y < SCALE; y++) {
  // a small 0.95 scale factor seems to add stability
  dsum += err[y] * 0.95;
}
// update this DAC and phase
dacs[phase] = int(dsum + dither + in/SCALE);
err[phase] = ((in/SCALE) − dacs[phase]);
// check to make sure DAC is only a fraction of the full 10-bit range
if ((dacs[phase] > (1024 / SCALE)) ∥ (dacs[phase] < 0)) {
  fprintf(stderr, “Range Error: %d, %d\n”, phase, dacs[phase]);
}
// sum all dacs, and print output
// repeat output to show images
sum = −512; // remove DC offset
for (y = 0; y < SCALE; y++) {
  sum += dacs[y];
}
printf(“%d\n”, sum);
if (SCALE == 1) {
  printf(“%d\n”, sum);
  printf(“%d\n”, sum);
  printf(“%d\n”, sum);
}
// increment phase
phase++;
if (phase >= SCALE) {phase = 0;}
}
}

Accordingly, the present invention may be realized in hardware, software, or a combination of hardware and software. The present invention may be realized in a centralized fashion in at least one computer system, or in a distributed fashion where different elements are spread across several interconnected computer systems. Any kind of computer system or other apparatus adapted for carrying out the methods described herein is suited. A typical combination of hardware and software may be a general-purpose computer system with a computer program that, when being loaded and executed, controls the computer system such that it carries out the methods described herein.

The present invention may also be embedded in a computer program product, which comprises all the features enabling the implementation of the methods described herein, and which when loaded in a computer system is able to carry out these methods. Computer program in the present context means any expression, in any language, code or notation, of a set of instructions intended to cause a system having an information processing capability to perform a particular function either directly or after either or both of the following: a) conversion to another language, code or notation; b) reproduction in a different material form.

While the present invention has been described with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the present invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the present invention without departing from its scope. Therefore, it is intended that the present invention not be limited to the particular embodiment disclosed, but that the present invention will include all embodiments falling within the scope of the appended claims.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US4415914 *Jun 30, 1981Nov 15, 1983International Business Machines CorporationGray scale electrostatic recording system and a stylus driver therefor
US5684482 *Mar 6, 1996Nov 4, 1997Ian A. GaltonSpectral shaping of circuit errors in digital-to-analog converters
US5748126 *Mar 8, 1996May 5, 1998S3 IncorporatedSigma-delta digital-to-analog conversion system and process through reconstruction and resampling
US6150970 *Mar 1, 1999Nov 21, 2000Sony CorporationMethod and system for digital to analog conversion
US6476746 *Nov 29, 2000Nov 5, 2002Texas Instruments IncorporatedCellular base station having a high speed, high resolution digital-to-analog converter with off-line sigma delta conversion and storage
US6489908 *Apr 30, 2001Dec 3, 2002Texas Instruments IncorporatedWireless local loop terminal and system having high speed, high resolution, digital-to-analog converter with off-line sigma-delta conversion and storage
US6535154 *Nov 5, 2001Mar 18, 2003Texas Instruments IncorporatedEnhanced noise-shaped quasi-dynamic-element-matching technique
US6603418 *Dec 14, 2001Aug 5, 2003Teradyne, Inc.High precision, high-speed signal source
US6603804 *Oct 1, 1999Aug 5, 2003Agere Systems Inc.Upsampling filter having one-bit multipliers for multiple spread-data streams
US6727832 *Nov 27, 2002Apr 27, 2004Cirrus Logic, Inc.Data converters with digitally filtered pulse width modulation output stages and methods and systems using the same
US6737999 *Sep 3, 2002May 18, 2004Analog Devices, Inc.Mismatch-shaping for a quadrature digital-to-analog converter
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US7501969May 29, 2007Mar 10, 2009Broadcom CorporationMethod and apparatus for an image canceling digital-to-analog converter (DAC)
US8098718 *Dec 17, 2009Jan 17, 2012Qualcomm IncorporatedApparatus and methods for digital-to-analog conversion with vector quantization
US8565707 *Oct 13, 2011Oct 22, 2013Qualcomm IncorporatedCancellation of spectral images in communication devices
Classifications
U.S. Classification341/144, 341/143
International ClassificationH03M1/66, H03M3/04
Cooperative ClassificationH03M3/50, H03M3/368
European ClassificationH03M3/30
Legal Events
DateCodeEventDescription
Oct 4, 2013FPAYFee payment
Year of fee payment: 8
Sep 4, 2013FPAYFee payment
Year of fee payment: 8
Sep 16, 2009FPAYFee payment
Year of fee payment: 4
Oct 26, 2004ASAssignment
Owner name: BROADCOM CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SCHONER, BRIAN F.;REEL/FRAME:015296/0215
Effective date: 20040818